1

Money supply and economic activity in South Africa – the relationship

updated to 2010

Introduction

In the eighties and early nineties a number of attempts were made to measure the

relationship between various measures of money supply growth and the growth in GDP,

GDE, Household Consumption Expenditure and Consumer Prices between 1966 and

1993. (82, 1984,1989, 1990b, 1993) This earlier work on the relationship between

money economic activity and prices was concluded in 1990 with an attempt to separate

monetary causes and effects. That is to estimate whether the money to expenditure and

income link was stronger than the income to money link, given the accommodative

nature of money supply responses. It was reported that the money to income link was

stronger than the reverse income to money influence, using a vector auto-regression

approach. (1990a). The purpose of this paper is to update this analysis to include the

past twenty years of data to establish whether or not money still matters for the SA

economy in the way it did.

Earlier conclusions and implications

Our earlier calculations had demonstrated that the growth rates of the narrower

measures of money were more highly correlated with the growth in GDP and especially

of Gross Domestic Expenditure (GDE) and Household Consumption spending than with

wider measures of money growth. The linkages between money supply growth and

inflation was found to be a statistically weak. It was explained that, given the openness

of the SA economy to exports and imports, the money supply would affect aggregate

demand or GDE much more directly than GDP. It was also understood that changes in

the exchange rate independently of domestic demand (exchange rate shocks) could

influence the trend in consumer prices.

The intention was to provide a complete analysis of the extent to which published

Reserve Bank monetary measures correlated with Reserve Bank measures of economic

activity. It was argued that there was little empirical evidence for the Reserve Bank view

that broader measures of the money supply (M3) would be a superior target to

narrower definitions (M0) for the purposes of monetary policy. The Reserve Bank had

pronounced in 1988 that

“….As a practical matter, movements in M0 over the past several years have been found

to correlate relatively poorly (more so than other monetary aggregates) with

movements in macro-economic variables such as nominal gross domestic product or the

2

general price level…” (SARB Quarterly Bulletin, March 1988 p.16-17) as quoted in Barr

and Kantor (1989 p. 292).

It was well appreciated in this body of work that a reduced form econometric approach

was inappropriate for addressing the issue of the endogeneity of the supply of money in

the South African context. It was understood that the operating procedures of the SA

Reserve Bank accommodated the demands for cash by the public and the banks at the

policy determined discount or repo rate and that this could make the supply of central

bank money and the money supply broadly defined as more the effect of than the cause

of economic activity.

In an analysis of the the De Kock Commission Report (1986) a full model of the money

supply process was set out and solved for the equilibrium overdraft rate and the broad

money supply. It was demonstrated that were it the official intention to achieve money

supply objectives with interest rate settings, the authorities would have to estimate

accurately the income elasticity of the demand for money, but also the many other

elasticities identified in the two reduced form equations - a clearly formidable task.

Some monetary history in charts and tables

The highly variable and pro-cyclical behaviour of the money supply and the high rates of

inflation over the extended period from 1966 to 2010 would seem to confirm the

practical difficulties faced by the SA Reserve Bank in the attempts it presumably made

to control the supply of money and bank credit and moderate the money and credit

cycles with interest rate settings. 1

In the figure below we show the long history of money supply and credit growth using

monthly data. These growth rates are compared in the figure to the annual change in

the co-inciding business cycle indicator published by the Reserve Bank to make the pro-

cyclicality of money and credit growth apparent. It will also be apparent that the growth

in money and credit and its cyclicality have remained highly elevated since 2000.

1 See for a recent explanation of the operating procedures of the Reserve Bank, N Brink and M Kock, Central bank balance sheet policy in South Africa and its implications for money-market liquidity, South African Reserve Bank Working Paper,

WP/10/01, December 2009

3

Fig 1. Annual growth in M3, the Note Issue and the Business Cycle in SA

(Monthly data 1971-2010)

-30

-20

-10

0

10

20

30

40

70 75 80 85 90 95 00 05 10

Growth in M3 Growth in Bank Credit Growth in Co-inciding Indicator

Source; SA Reserve Bank Data Base

This impression of rapid and highly variable money and credit supply growth is

confirmed by the summary statistics provided in the Tables below. Yet while the

generally rapid M3 and bank credit growth rates are very similar before and after 2000,

two differences in the observed outcomes before and after 2000 should be noted. Firstly

the growth in the note issue and the growth in the broader measure of money (M3)

while very similar on average before 2000, diverge thereafter as the average growth in

the note issue slowed down after the turn of the century. Secondly the average head

line inflation rate after 2000 is significantly lower than before. Inflation targeting applied

in SA after 2000 may therefore be regarded as successful in helping to reduce the

average rate of inflation.

Yet it should be noted that while the average rate of inflation declined significantly after

2000, the inflation rate has remained very volatile. Indeed inflation became significantly

more volatile after 2000 than in the earlier period. The standard deviation of the

inflation rate was 3.06 % between January 2000 and December 2010. While the inflation

4

rate was much higher between 1971 and 1999 averaging 11.61% the Standard Deviation

of the inflation rate was a similar 3.69% Thus the co-efficient of variation of inflation (the

standard deviation divided by the lower mean) was 0.52 for the period 2000-2010 and

much lower 0.32 on average between 1971 and 1999. We provide a full explanation of

the divergent money supply trends below.

Table 1; Money supply growth and inflation 1971 -2010. 2

Sample: 1971:01 2010:12

M3 NOTE ISSUE BANK CREDIT INFLATION

Mean 15.00 13.23 15.46 10.00

Median 15.10 11.75 15.48 10.04

Maximum 29.34 33.31 35.88 20.94

Minimum 0.14 1.61 -1.09 0.13

Std. Dev. 5.75 5.86 6.59 4.35

Observations 468.00 468.00 468.00 468.00 Sample: 2000:01 2010:12

M3 NOTE ISSUE BANK CREDIT INFLATION

Mean 14.71 10.21 13.64 5.92

Median 15.03 9.91 12.75 5.77

Maximum 27.25 17.87 27.49 13.06

Minimum 0.14 2.90 -1.09 0.13

Std. Dev. 6.64 3.17 7.77 3.06

Observations 132 132 132 132 Sample: 1971:01 1999:12

M3 NOTE ISSUE BANK CREDIT INFLATION

Mean 15.11 14.42 16.18 11.61

Median 15.12 13.11 15.70 11.78

Maximum 29.34 33.31 35.88 20.94

Minimum 1.93 1.61 5.63 1.71

Std. Dev. 5.37 6.24 5.92 3.69

Observations 336.00 336.00 336.00 336.00 Source; SA Reserve Bank Data Base

2 Calculated monthly as annual year on year growth rates in money and the

Consumer Price Index (CPI) The growth rates of M3 and the Note issue are

compound growth rates Log/(x(x(-12))/.01 while headline inflation is calculated

conventionally as a simple percentage (CPI-CPI(-12))/CPI(-12)/.01

5

In the tables below we summarize the results of the updated analysis that takes into

account the full history of the relationship between alternative measures of money

supply growth and economic activity and inflation between 1967 and 2010. The results

are presented below in Table(2) and Table(3). Results are also presented for the period

1967-1981 to correspond with the earlier studies and for the periods 1967-1999 and for

the period 2000-2010.

The measures of interest are the extent to which various monetary measures ranging

from the narrowest monetary measure to the broadest monetary measure can be used

to fit a range of measures of economic activity. Since the measures of economic activity

are only available quarterly, all variables are considered at quarterly intervals. To render

the variables stationary both the measure of economic activity and the measure of the

monetary aggregate are computed in growth rate form and to remove any seasonality

these growth rates are computed as year-on-year growth rates.

.

The model is of the form

4

01

% * %i t ii

E M

Where: 4% 100* ( / )t e t tE Log E E

4% 100* ( / )t e t tM Log M M

For the monetary measures we use M1, M2, M3 and notes in circulation. For the

measures of economic activity we use GDP, GDE and Household Consumption. We also

include a measure of inflation (based on the consumer price index). The tables give the

summary statistics for R2 (adj. dof), F and the Durbin-Watson (DW) statistics for ordinary

least squares regressions for the indicated periods for the model estimations. The R2

(adj. dof) and F statistics are parallel measures of model fit and give, respectively, a

measure of the (least-squares) degree of fit and the overall statistical significance of the

monetary aggregates in explaining the economic activity variable. The DW gives a

measure of the first-order autocorrelation of the residuals.

6

It should be noted that while computing these economic activity and monetary

measures in growth rate form is necessary to render the variables stationary and thus

remove any possibility of spurious correlations arising, such growth rates

transformations will tend to induce some autocorrelation in the residuals. Regressions

between smooth variables will generally produce autocorrelated residuals and these

may lead to difficulties in interpretation. Thus in this analysis we give tables for the

regressions both with and without an autocorrelation adjustment. Note also that this

study is looking primarily at the relative levels of explanation of measures of economic

activity by different measures of money across different time periods and as such it is

the (relative) overall measures of fit that are important rather than the estimates of the

individual beta coefficients in the model.

7

Table 2. Money and Economic Activity 1967.1-2010.3 (Quarterly data) without an autocorrelation adjustment M1G M1G M1G M1G M2G M2G M2G M2G M3G M3G M3G M3G NOTESG NOTESG NOTESG NOTESG 67.1–81.3 67.1–99.4 ‘00.1–‘10.3 67.1–‘10.3 67.1–81.3 67.1–99.4 ‘00.1–‘10.3 67.1–‘10.3 67.1–81.3 67.1–99.4 ‘00.1–‘10.3 67.1–‘10.3 67.1–81.3 67.1–99.4 ‘00.1–‘10.3 67.1–‘10.3

GDPG R2-bar 0.280 0.028 0.643 0.073 0.209 0.121 0.134 0.144 0.330 0.209 0.257 0.175 0.533 0.337 0.009 0.328 D.W. 0.549 0.394 1.150 0.386 0.478 0.456 0.721 0.449 0.550 0.509 0.835 0.458 0.656 0.597 0.553 0.564 F-stat 8.524 2.250 26.255 5.587 6.112 6.986 3.169 10.756 10.517 12.560 5.851 13.327 26.083 24.561 1.123 30.631

GDEG R2-bar 0.555 0.087 0.627 0.138 0.460 0.228 0.242 0.228 0.514 0.264 0.319 0.234 0.753 0.544 0.196 0.522 D.W. 0.782 0.439 0.947 0.445 0.627 0.502 0.712 0.507 0.646 0.526 0.754 0.498 1.194 0.871 0.587 0.816 F-stat 25.098 5.181 24.552 10.301 17.478 12.538 5.462 18.093 21.488 16.685 7.549 18.748 67.903 56.324 4.411 67.357

HHG R2-bar 0.673 0.086 0.686 0.160 0.584 0.263 0.255 0.273 0.614 0.327 0.428 0.222 0.740 0.573 0.241 0.572 D.W. 0.927 0.284 0.383 0.255 0.683 0.362 0.465 0.324 0.665 0.380 0.486 0.278 0.856 0.596 0.334 0.535 F-stat 40.704 5.101 31.638 12.019 28.162 16.606 5.799 22.802 31.799 22.233 11.482 17.510 63.741 63.053 5.452 82.104

CPIG R2-bar 0.283 0.038 0.380 0.091 0.530 0.261 0.120 0.239 0.650 0.253 0.275 0.159 0.405 0.326 0.195 0.293 D.W. 0.178 0.115 0.342 0.108 0.345 0.166 0.238 0.147 0.516 0.179 0.302 0.126 0.228 0.177 0.348 0.167 F-stat 8.640 2.727 9.573 6.773 22.797 16.394 2.902 19.253 36.894 15.784 6.314 11.974 15.995 23.366 4.400 26.097

Table 3. Money and Economic Activity 1967.1-2010.3 (Quarterly data) with an autocorrelation adjustment M1G M1G M1G M1G M2G M2G M2G M2G M3G M3G M3G M3G NOTESG NOTESG NOTESG NOTESG 67.1–81.3 67.1–99.4 ‘00.1–‘10.3 67.1–‘10.3 67.1–81.3 67.1–99.4 ‘00.1–‘10.3 67.1–‘10.3 67.1–81.3 67.1–99.4 ‘00.1–‘10.3 67.1–‘10.3 67.1–81.3 67.1–99.4 ‘00.1–‘10.3 67.1–‘10.3

GDPG R2-bar 0.803 0.728 0.800 0.749 0.794 0.732 0.755 0.749 0.801 0.736 0.755 0.752 0.813 0.746 0.705 0.756 D.W. 2.123 2.063 2.116 2.067 2.141 2.092 1.843 2.079 2.050 2.086 1.833 2.109 2.373 2.131 1.835 2.113 F-stat 47.429 70.698 34.635 104.397 44.980 72.033 26.925 104.470 47.006 73.494 26.831 105.958 58.497 82.608 21.076 113.574

GDEG R2-bar 0.816 0.711 0.761 0.723 0.792 0.724 0.746 0.730 0.803 0.730 0.744 0.731 0.848 0.757 0.719 0.761 D.W. 2.289 2.318 2.295 2.320 2.410 2.371 1.869 2.330 2.473 2.387 1.779 2.363 2.099 2.298 1.961 2.272 F-stat 51.535 65.024 27.740 91.486 44.313 69.061 25.617 94.781 47.354 71.329 25.423 95.219 74.759 87.795 22.440 116.809

HHG R2-bar 0.828 0.805 0.927 0.839 0.808 0.806 0.872 0.838 0.819 0.811 0.885 0.841 0.847 0.837 0.867 0.856 D.W. 2.249 2.090 2.034 2.046 2.260 2.108 1.229 2.041 2.170 2.151 1.242 2.078 2.361 2.230 1.306 2.165 F-stat 56.035 108.153 107.842 181.772 48.878 109.018 58.091 180.142 52.699 112.737 65.917 184.572 74.037 144.276 55.786 217.412

CPIG R2-bar 0.945 0.922 0.905 0.932 0.945 0.925 0.898 0.934 0.941 0.923 0.905 0.934 0.947 0.934 0.878 0.935 D.W. 2.643 1.708 0.954 1.476 2.708 1.737 0.756 1.505 2.558 1.732 0.849 1.541 2.366 1.635 0.678 1.421 F-stat 195.122 306.925 81.164 472.767 197.004 320.440 74.728 490.971 183.491 314.727 81.090 493.279 237.490 393.575 61.590 528.315

8

The result of the exercise confirms the consistency of the relationship between money,

economic activity and prices in South Africa over the extended period 1966 - 2010. The

results are very much in line with those presented before. The closest statistical

associations are found between growth in the narrow money supply and growth in GDE

and Household Consumption Expenditure. While the measures of money supply for the

earlier period are largely unchanged, the national income statistics and of prices have

been revised, sometimes significantly, giving rise to somewhat different estimates from

those obtained in the earlier published estimates for the same overlapping periods.

It should be noted that the results that adjust for autocorrelation have much higher fits

than those which do not adjust for autocorrelation. This is because the model with

autocorrelated fits is effectively using the lagged dependent variable are an additional

explanatory variable. Thus, for example, in the case of GDP growth rates, the model fits

against growth rates of the monetary measures is effectively also using lagged growth

rates of GDP as an additional explanatory variable. It can thus be argued that the fits,

without autocorrelation adjustment, provide a clearer comparative standard for

measuring the various relative magnitudes of the effects of the different monetary

measures across different periods of time.

Focusing on the results without the autocorrelation adjustment, it is seen that the

money measures fit economic activity very satisfactorily over the extended period 67-

2010, with the best fits to be obtained from the narrowest measure of the money

supply, the note issue. As was noted in the earlier work, the relationship between the

growth in money and expenditure, GDE or Household Consumption, has remained

consistently more significant than that between the growth in money supply and GDP.

The links between money and inflation have remained statistically rather weak ones, as

may be seen. The trade balance and independent exchange rate movements have

continued to interrupt the money supply to aggregate expenditure and in turn output

and price level linkages. The best statistical fits for the relationship between narrow

money and expenditure and output as measured by R squared and the F statistic, are

found over the period 1967 to 1981, the focus of the original studies, as may be seen.

The goodness of fit and significance of the coefficients of the models is not quite as

strong for the period 1967-1999 a period that includes the economically very turbulent

and inflationary mid eighties and early nineties.

One important difference in the estimates should however be recognised. That is for the

sub-period ‘00 -’10, the growth in the broader measures of money, for example M3,

became somewhat better related to measures of economic activity than did the growth

in the narrowly defined money supply and expenditure. Furthermore the statistical

relationship between the growth in the broad and narrow definitions of money and the

growth in expenditure became statistically less significant over the past ten years and

more.

9

It should however be recalled, as shown in the tables above, that the relationship

between growth in the note issue and the broader definitions of money changed over

the past ten years. Growth in narrow and broader money diverged more after ’00 than

before with the average growth in M3 and Bank Credit significantly exceeding that of the

growth in the note issue by an average four to five per cent per annum over this period.

The correlation between the annual growth in the note issue and M3 measured

quarterly also declined from 0.579 between 1966 and 1999 to 0.297 between ‘00 and

‘10. 3 (See figure below)

Fig 2. Growth in Note Issue and growth in M3 ‘00.1-’10.3

0

5

10

15

20

25

00 01 02 03 04 05 06 07 08 09 10

GM3 GNOTES

Source; SA Reserve Bank Data Base

3 Growth rates have been compound percentage rates of growth calculated quarterly as Log(x/x(-

4))/.01

10

Accordingly the money multiplier, measured as the ratio of the broadly defined money

supply M3 to the note issue increased from an average of about 16 times between 1966

and 1999 to about 28 times by ‘08. It would appear that this multiplier has stabilised at

about the 28 times. (See below)

Fig.3 The money multiplier; Ratio of M3 to Note Issue 1966-’10

(Monthly Data)

12

16

20

24

28

32

70 75 80 85 90 95 00 05 10

M3/Note Issue

Source; SA Reserve Bank Data Base

Explaining the increase in the money multiplier

The SA banks have held and continue to hold minimal amounts of reserves in excess of

their required cash reserve holdings. They rather consistently borrow cash from the

Reserve Bank to satisfy reserve requirements. Therefore in SA, M0 adjusted for reserve

requirements becomes equivalent to the note issue. The Federal Reserve Bank of St

Louis, the originator of the St Louis equations designed to identify how much money

mattered for the US, defines the US Money Base as the sum of Notes plus cash reserves

held by the commercial banks with the Federal Reserve system, less their required

reserves.4

4 See Richard G.Anderson and Kenneth A Kavajecz, A Historical Perspective on the Federal Reserve’s

Monetary Aggregates: Definition, Construction and Targeting. http://research.stlouisfed.org/aggreg/

11

An explanation for this structural change after ‘00 is therefore called for. The reason for

the slower growth in the supply of notes compared to the growth in M3 and in bank

credit can be found in the reduced demand for notes exercised by the banks themselves.

This reduced demand for notes followed a decision taken by the Reserve Bank in ‘02 not

to allow the notes held by the banks to qualify as required cash reserves. The banks

were allowed to phase in the replacement of deposits at the Reserve Bank for cash in

their tills ATM’s and vaults between ‘02 and ‘04.5

The policy of the Reserve Bank is to meet the demand for notes as expressed by the

public and the banks. As mentioned previously the Bank has never adopted any explicit

targets for the money supply or the money base. A slower rate of increase in demands

for notes by either the public or banks would lead automatically mean a slower rate of

increase in the supply of cash to the system.

Using the model of the money supply process, we can show below how a reduction in

the demand for notes by the banking system, other things remaining the same, will lead

to an increase in the money multiplier,that is in the ratio M3/Note issue or M3/Money

Base. It may be seen below that the ratio of notes held by the banks to their deposit

liabilities declined significantly after ‘02 while the ratio of their cash reserves held as

deposits with the Reserve Bank moved strongly in the opposite direction as we show

below. The ratio of notes to deposits issued by the banks declined from about one and

half per cent of their total deposits issued in ‘02 to about half of one per cent by ‘10. The

ratio of the deposits held by the banks at the Reserve Bank increased from less than 1

per cent in ‘02 to about 2.5% by late ‘10. The share of the note issue held by the banks

declined over the same period from about 25% of the issue to about 15%. Over the

same period, as mentioned, the money multiplier that is the ratio of M3/Note Issue

increased from about 17 times in ‘00 to about 28 times by ‘08 where after the money

multiplier appears to have stabilised and declined moderately.

It is not clear that such an increase in the money multiplier was anticipated when the

regulation to exclude notes from classification as required reserves was introduced. Nor

is it clear that such an increase, even if anticipated, would have been of concern to the

Reserve Bank, given its modus operandi to accommodate, rather than to attempt to

control directly the notes or deposits it supplies to the system.

It is made clear with the aid of the money supply model presented below that the banks

could meet the demand for bank deposits and the demand for bank credit that grew so

strongly between ‘02 and ‘08, in part by reducing their real demand for notes as an

5 Brink and Kok in their paper (2009) note that “The last change to the cash reserve ratio was made in 2001, when the qualifying of vault cash as part of banks’ cash reserves was phased out over a period of four years” No explanation for this change in policy was provided in the paper.

12

alternative to increasing their demands for cash reserves held as deposits with the

Reserve Bank.

13

Fig 4. SA Banks; Ratio of notes held and deposits at Reserve Bank to

Deposit Liabilities.

0.000

0.005

0.010

0.015

0.020

0.025

0.030

86 88 90 92 94 96 98 00 02 04 06 08 10

Notes held/Deposit Liabilities Reserve Bank Deposits/Liab

Source; SA Reserve Bank Data Base

14

Fig 5.Ratio of Notes held by Banks to Total Note Issue

0.10

0.15

0.20

0.25

0.30

0.35

86 88 90 92 94 96 98 00 02 04 06 08 10

Notes held by banks/Notes

Source; SA Reserve Bank Data Base

Explaining the money supply process in South Africa- with an emphasis on the role of

the demand for notes

The model of the money supply process presented below and the derivation of the

money multiplier (equation 15) will indicate the important role played by the ratio of

notes held by banks to their deposit liabilities. It can be easily seen in equation 15 that a

large decline in the ratio of notes held by the banks (nb) will be associated with an

15

increase in the money supply, broadly defined, and in the money multiplier, the ratio of

the money supply, broadly defined, to the money base. (M/MB) (see equations 16 to 19)

16

A model of the money supply process in South Africa

17

A number of observations about this money supply model can be made. Firstly, as

mentioned before, it has not been the practice of the SA commercial banks to hold

significant excess cash reserves. Unlike their US counterparts after the Global Financial

Crisis, SA banks have continued to hold minimum amounts of cash in the form of

deposits at the Reserve Bank in excess of the legal requirement of them to satisfy their

18

cash to deposit ratios. SA banks however consistently borrow cash from the Reserve

Bank. Thus free reserves as per equation 6 in the model, the difference between excess

and borrowed reserves, are consistently negative. (see figure below)

Fig 6. Composition of SA Bank Reserves (R million)6

0

10000

20000

30000

40000

50000

60000

99 00 01 02 03 04 05 06 07 08 09 10

Reqd Reserves Borrowed Reserves Excess Reserves

Required

Borrowed

Excess

R m

illions

Source; SA Reserve Bank Data Base

This dependence (at the margin) of the commercial banks on cash supplied by the

Reserve Bank is intentional. It helps make the central bank’s discount rate, or in modern

terminology, its repurchase or the repo rate, effectively the bench mark short term rate

of interest in the money market. Therefore because almost all of the deposits held by

the banks at the Reserve Bank, its cash reserves, are required reserves (equations 4 and

5) the money base in South Africa or M0, (equation 1) if adjusted for reserve

requirements then reduces to the Note Issue. The so called high powered money in the

system, the quantity that is powered up to M3 is then Note Issue rather than M0 that

consists of the Note Issue plus the cash reserves of the banks held with the Reserve bank

that are consistently almost entirely required reserve holdings.

6 Note on data sources. Required Reserve Balances, Reserve Bank Data Bank Series KBP1014m,

Borrowed Reserves, Series KBP1025m, Excess Reserve Series KBP1013m. The statistics for excess and

borrowed reserve date back only to March 1998.

19

This adjustment for required reserves is consistent with monetary theory that regards

the excess supply of money over the demand to hold money, not the money supply

itself, as influencing aggregate demand and so the price level. Or in other words excess

cash reserves held by the banks matter for the economy because they may lead to more

bank lending and an increase in the money supply via the deposit money multipliers.

Required reserves are in effect frozen on the books of the commercial banks.

Conclusion

It will be clear from the tables and figures shown above that broadly defined money

supply growth and the growth in the supply of bank credit has remained as highly

variable and as highly pro cyclical since ‘00 as it was before. The era of inflation targeting

has not brought less variable money supply and bank credit supply growth. It has

however brought lower inflation on average but not less variable inflation, especially if

the co-efficient of variation of the inflation rate is the appropriate measure of volatility.

The ability of the SA Reserve Bank to moderate the money and bank credit cycles,

utilising interest rates as the primary instrument of monetary policy, seems as elusive as

ever. The supply of money and credit in South Africa appears to respond primarily and

endogenously to demands for cash and credit, given interest rate settings. Adjustments

to these policy determined interest rates appear to lag well behind the demands for

extra credit and the cash to satisfy such demands in both directions.

Or in other words policy determined interest rates are set too low to restrain the money

and credit supplies when the money and credit cycles have gained momentum and then

remain too high when demands for credit and money slow down to prevent the growth

in money and credit from slowing down precipitately.

It would appear that the surge in money and credit growth between ‘03 and ‘08 was

accommodated by an unexpected reduction in demands for cash by the banks rather

than, as would be more usual, by an increase in the supply of cash reserves made

available by the Reserve Bank to the banks. The apparent unpredictability of the

demands by the banks for notes, in the face of changes in the composition of qualifying

cash reserves, represents an additional complication when setting interest rates to be

(hopefully) consistent with appropriate money and credit supply objectives.

The evidence in the form of still highly variable and pro-cyclical money and bank credit

cycles suggests that the task of stabilising the money and credit cycle, utilising interest

rate settings, remains as elusive as ever and beyond the capacity of monetary policy in

South Africa. The operating procedures of the Reserve Bank continue to prove incapable

of effectively moderating, in an effectively contra cyclical way, the money and credit

cycles.

20

References

Anderson, Richard G and Kavajecz, Kenneth A, (2011) A Historical Perspective on the

Federal Reserve’s Monetary Aggregates: Definition Construction and Targeting.

http://research.stlouisfed.org/aggreg/

Barr, G.D.I. and Dietzsch, C.H. (1980). The direction of causality between money and

income - the South African case. Studies in Economics and Econometrics, 7, 24-29.

Barr, G.D.I. and Kantor, B.S. (1982). Money and economic activity: Some alternative

results. South African Journal of Economics, 50, 375-377.

Barr, G.D.I. and Kantor, B.S. (1984). Interests rates, exchange rates and the money

supply in South Africa. The Investment Analysts Journal, 23, 45-50.

Barr, G.D.I. and Kantor, B.S. (1989) Money and economic activity - A response to the

Reserve Bank, South African Journal of Economics, 57(3), 292-298.

Barr, G.D.I. and Kantor, .S. (1990) The Application of a vector autoregressive model to

money income and price links in the S.A. Economy, Journals for Studies in Economics and

Econometrics, 14(1), 39-49.

Barr, G.D.I. and Kantor, B.S. (1990) Money and economic activity, a response to the

Reserve Bank: a reply. South African Journal of Economics, 58(3), 368-369.

Barr, G.D.I. (1990) A vector autoregressive model - reply. Journal for Studies in

Economics and Econometrics, 14(3), 89.

Barr, G.D.I. and Kantor, B.S. (1993) The Money-Income Causality Debate in South Africa:

Comment. South African Journal of Economics, 61(1), 84-86.

Barr, G.D.I & Kantor, B.S. (2002), The South African Economy and its Asset Markets: An

Integrated Approach, SAJE, 70, 1, 53-77.

Brink N & Kok (2009), Central bank balance sheet policy in South Africa and its

implications for money-market liquidity, South African Reserve Bank Working Paper,

WP/10/01, December 2009